Navigate back to the homepage


beyond excelsior

Software development, application and system architecture

about meHow it's doneContact
Link to $ to $ to $ to $ to $ to $ to $

Azure Functions with typescript

Luke Celitan
category: post, reading time: 4 min

Azure Functions with typescript


The so-called “Serverless” technology was introduced by AWS with their Lambda service, and Azure quickly followed suit with their own rendition, Azure Functions. Azure Functions is an excellent way to run small pieces of code in the cloud without having to worry about the underlying infrastructure. Here is a quick guide on how to create Azure Functions using TypeScript.

Azure Functions v4

Azure Functions v4 is the latest version of Azure Functions. In addition to many languages supported by Azure Functions including C#, JavaScript, Java, Python, and PowerShell, Azure Functions v4 has added support for TypeScript.

The almost native support for TypeScript in Azure Functions v4 is a great addition, as it’s included in the development runtime, CLI commands, and templates. In the JavaScript version, users have to add those cumbersome binding JSONs, but this isn’t necessary in the TypeScript version. Unfortunately, the underlying Func engine (Azure Core Tools) does not support custom path aliases (@utils) in the tsconfig.json, a feature beloved by TypeScript developers.

The Serverless use cases

Back in the day, when serverless was starting to gain serious hype, dreams of automatic upscaling went wild. Despite Azure Functions clearly defining its purpose, some people started using them as a replacement for traditional web servers. It became quickly evident that the costs for functions were higher than containers or VMs.

  • Azure Functions are exceptional for small pieces of code that are triggered by events.
  • They’re also an excellent solution for moving interfaces and functions that are significant bottlenecks for the main application.
  • They can circumvent current limitations of the main application by creating a new function that performs tasks incompatible with the main server, or when changes to the system are costly or risky.


We are obviously working with Node.js here, making the development fairly straightforward. The func CLI tool, part of the Azure Core Tools, is used to create the functions.

The actual SDK package is @azure/functions, which is used to create the functions and handle requests.

The Node.js “programming model” should not be confused with the Azure Functions “runtime”. The programming model defines how you author your code and is specific to JavaScript/TypeScript. The runtime defines the underlying behavior of Azure Functions and is shared across all languages. The programming model version is tied firmly to the version of the @azure/functions npm package, and is versioned independently from the runtime.

Unfortunately, both the runtime and the programming model are on “4” as their latest major version, but this is purely coincidental. You cannot mix the v3 and v4 programming models in the same function app. As soon as you register one v4 function in your app, any v3 functions registered in function.json files are ignored.

Chart to image function

My Azure Function example/template is an HTTP trigger function that creates a PNG based on the node-canvas package given specific request parameters.

Recently, I was tasked with proving a concept for a chart to image conversion. My main application’s current stack was having a hard time accepting Canvas-node, known for its difficult installation*. Azure Function seemed like an elegant solution for this problem.

*This is a subjective opinion based on the node-canvas issues on the GitHub repo.

As usual, I added much-needed development tools such as ESLint, Prettier, Husky, and Commitlint for proper code quality maintenance. Additionally, I added the Vitest test for unit and integration testing.

You can see a full github repo example here

1import {
2 app,
3 HttpRequest,
4 HttpResponseInit,
5 InvocationContext,
6} from '@azure/functions';
8import { extractColors, createCanvas } from '../utils';
9import { barChartOptions } from '../charts';
10import { Chart, ChartConfiguration, ChartItem, Plugin } from 'chart.js/auto';
12export async function barChart(
13 request: HttpRequest,
14 context: InvocationContext,
15): Promise<HttpResponseInit> {
16 try {
17 context.log(`Http function processed request for url "${request.url}"`);
19 const { canvas, ctx } = createCanvas(400, 400);
21 const plugin: Plugin = {
22 id: 'customCanvasBackgroundColor',
23 beforeDraw: (chart, _args, options) => {
24 const { ctx } = chart;
26 ctx.globalCompositeOperation = 'destination-over';
27 ctx.fillStyle = options.color || '#99ffff';
28 ctx.fillRect(0, 0, chart.width, chart.height);
29 ctx.restore();
30 },
31 };
33 const plugins = [plugin];
35 const labels = request.query.get('labels')?.split(',');
36 const dataValues = request.query.get('data')?.split(',').map(Number);
38 if (
39 !labels ||
40 labels.length === 0 ||
41 !dataValues ||
42 dataValues.length === 0
43 ) {
44 throw new Error('Unable to create');
45 }
47 const backgroundColorsRaw = request.query.get('backgroundColors');
48 const borderColorsRaw = request.query.get('borderColors');
50 const backgroundColors = extractColors(backgroundColorsRaw);
51 const borderColors = extractColors(borderColorsRaw);
52 const data = {
53 labels: labels,
54 datasets: [
55 {
56 label: 'My First Dataset',
57 data: dataValues,
58 backgroundColor: backgroundColors,
59 borderColor: borderColors,
60 borderWidth: 1,
61 },
62 ],
63 };
65 const chartConfig = {
66 type: 'bar',
67 data: data,
68 options: barChartOptions,
69 plugins,
70 };
72 new Chart(ctx as unknown as ChartItem, chartConfig as ChartConfiguration);
74 const pngBuffer = canvas.toBuffer('image/png');
76 // Set the headers and send the PNG buffer in the response
77 request.headers.set('Content-Type', 'image/png');
78 request.headers.set('Content-Length', pngBuffer.length.toString());
79 return {
80 status: 200,
81 body: pngBuffer,
82 headers: request.headers,
83 };
84 } catch (error) {
85 context.log(error);
86 return {
87 status: 500,
88 body: 'Internal server error',
89 };
90 }
93app.http('barChart', {
94 methods: ['GET', 'POST'],
95 authLevel: 'anonymous',
96 handler: barChart,

This function expects almost all data to come in the query parameters, and it dynamically creates the bar chart image. When calling the function, labels, data, background, and border colors must be supplied.


The final result is a simple bar chart image that can be downloaded or used in applications, emails, or other mediums that might work better with images.

Space chart


In summary, serverless functions like Azure Functions have surfaced as innovative solutions that offer a different perspective to addressing common problems in scaling and software compatibility. As showcased by the example above, serverless functions can tackle specific tasks that might not be feasible or efficient within a primary application’s context.

The advent of AWS’s Lambda service marked the beginning of a new era of serverless computing, and the technology was soon embraced by Azure in the form of Azure Functions. This new technology was appealing as it allowed small code pieces to be executed in the cloud without worrying about managing the underlying infrastructure. Whatever Azure lacks in first-mover advantage, it tries to compensate with new features and support for a wide array of languages, notably recently introducing TypeScript support.

However, serverless computing comes with its unique challenges and misconceptions. Despite their seemingly endless benefits, Azure Functions or its equivalents are not the ultimate solution to every problem—they cannot and should not replace traditional web servers entirely. It’s crucial to understand that even though serverless functions can auto-scale, their costs can exceed those of containers or VMs, which can be surprising and detrimental for uncritical adopters.

Instead, serverless functions excel when applied to small tasks, driven by events. They also function well when moving interfaces and tasks that weigh heavily on the primary application, allowing for improved overall performance. They can also serve as a perfect solution to address compatibility issues, circumventing limitations present within the primary application by offering a platform to build functions that can carry out tasks that the main server is incompatible with.

To conclude, serverless functions, along with exciting emerging technological advances in the industry, should be seen as beneficial tools within a developer’s toolbox. However, like any tool, their adoption should be based on needs, feasibility, and context. A careful understanding of their use cases, costs, advantages, and disadvantages is mandatory for making an informed choice about when and how to use Azure Functions or any serverless technology. It is essential to remember that advancements in technology inevitably come with their own set of challenges, and serverless functions are no exception. Therefore, they should be approached with caution, critical evaluation, and a readiness to adapt and learn.